From: Chris Frymann
Date: Wed, 25 Feb 2009 07:13:10 -0800
To: <dot.porter@gmail.com>
Subject: Re: Mapping from RDF to METS

Hi Dot,

Your question has been passed along to me so I will attempt a short answer.

First, you might try the Google search:

mapping from RDF to METS

which will return a number of useful references.

Second, and from a more abstract point of view, any RDF can be expressed in a number of different formats, including RDF/XML, and since METS, at one level, can act as a simple container or wrapper, any XML can be wrapped in or contained in a METS file, therefore RDF or RDF/XML can be embedded/mapped into a METS container. Simply putting RDF/XML into METS may or may not be something one wants to do, but it makes the point that there is no inherent limitation on getting RDF into a METS container.

Third, and a more practical example, in my environment we have developed a small MODS ontology which enables us to express MODS in RDF (or RDF/XML in particular). We then use a locally tailored XSL transform to automatically convert our RDF/XML into METS with a MODS section. By way of a little additional explanation, we have gone to the effort of expressing our metadata in RDF as well as METS for much the same reasons expressed in the following text copied from:

Semantic Web technologies for digital preservation : the SPAR project

Note particularly, the sentence:

“All the relevant metadata being available in the METS files and in the reference information, we had to map them and index them.”

Although related to your question, technically this is quite a different issue, as it address the topic of mapping METS to RDF rather than RDF to METS.

Here is the larger relevant excerpt from the SPAR article that makes a case for expressing metadata in RDF.

“When designing a system for the long term, it is not possible to imagine all the queries that will be relevant in the future.

Complex queries involve data formats, periods of time, events that have occured to a series of digital objects, software or human agents involved in the processes, etc. The flexibility of the data management is thus a key point in the development of SPAR and we had to take this into account when designing the indexation functions for the data management module.

All the relevant metadata being available in the METS files and in the reference information, we had to map them and index them. Four options were possible :

a XML database,
a relational database,
a RDF triple store or
a search engine.

A risk analysis taking into account implementation issues, functionnal opportunities and persistence in the long term, revealed the RDF triple store as the best candidate for managing metadata in this context :

* the mapping from XML to RDF was considered more relevant and evolutive than the mapping from XML to a relational database,

* the querying and access functionalities were richer than those provided by a search engine, thanks to expressiveness of a standardised query language SPARQL,

* the scalability and robustness was expected to be better than with a XML database, taking into account the amount of expected metadata to be handled by the system (2 billions triples after 2 or 3 years).

Regarding the latter, a benchmark was realized with Virtuoso3 and 2 billions triples were generated with the the LUBM4 ; the results of this prototype were satisfying and confirmed our choice.”

Please feel free to post my response to your group, if you like.

Chris Frymann (cfrymann@ucsd.edu)
Digital Library Architect

UC San Diego
>> Dear List,
>>
>> I’m writing to see if anyone here has thoughts and/or experience
>> mapping from RDF to METS. METS is of course much more rich than RDF,
>> but is it possible to create even a skeletal METS record (containing
>> only a file section and structural map) from RDF triples? Many thanks
>> for any thoughts or advice.
>>
>> Dot
>>
>> –
>> Dot Porter (MA, MSLS) Metadata Manager Digital Humanities
>> Observatory (RIA), Pembroke House, 28-32 Upper Pembroke Street,
>> Dublin 2, Ireland
>> — A Project of the Royal Irish Academy –
>> Phone: +353 1 234 2444 Fax: +353 1 234 2400 http://dho.ie
>> Email: dot.porter@gmail.com

Semantic Web technologies for digital preservation : the SPAR project

explains “…why RDF is relevant for digital preservation and how it will be implemented in SPAR”.

Offers the following case for why METS is not enough.

“All the relevant metadata being available in the METS files and in the reference information, we had to map them and index them. Four options were possible :

a XML database,
a relational database,
a RDF triple store or
a search engine.

A risk analysis taking into account implementation issues, functionnal opportunities and persistence in the long term, revealed the RDF triple store as the best candidate for managing metadata in this context :

* the mapping from XML to RDF was considered more relevant and evolutive than the mapping from XML to a relational database,

* the querying and access functionalities were richer than those provided by a search engine, thanks to expressiveness of a standardised query language SPARQL,

* the scalability and robustness was expected to be better than with a XML database, taking into account the amount of expected metadata to be handled by the system (2 billions triples after 2 or 3 years).

Regarding the latter, a benchmark was realized with Virtuoso3 and 2 billions triples were generated with the the LUBM4 ; the results of this prototype were satisfying and confirmed our choice.”